Structure of the Anion-Transport Protein of the ... - Semantic Scholar

Biochem. J. (1979) 181, 477-493 Printed in Great Britain

477

Structure of the Anion-Transport Protein of the Human Erythrocyte Membrane FURTHER STUDIES ON THE FRAGMENTS PRODUCED BY PROTEOLYTIC DIGESTION By David G. WILLIAMS, Robert E. JENKINS and Michael J. A. TANNER Department of Biochemistry, University ofBristol, Bristol BS8 1 TD, U.K.

(Received 6 February 1979) The topology of the human erythrocyte membrane anion-transport protein (band 3) has been investigated by isolation and peptide 'mapping' of the major and minor fragments derived from proteolytic cleavage of the lactoperoxidase 1251-labelled protein in erythrocytes and erythrocyte membranes. The content, in each fragment, of lactoperoxidase 1251I-labelled sites (which have a known location in the extracellular or cytoplasmic domain of the protein), together with the location of the sites of proteolytic cleavage yielding the fragments, has allowed us to determine the alignment of the fragments on the linear amino acid sequence and to infer the topology of the polypeptide in the membrane. The results suggest that a region in the C-terminal portion of the polypeptide forms part of the cytoplasmic domain of the protein in addition to a large N-terminal segment. The membrane-bound regions of the protein are located in the C-terminal two-thirds of the molecule. In this region the polypeptide chain traverses the membrane at least four times and an additional loop of polypeptide is either embedded in the membrane or also penetrates through it to the other surface. The location of the lectin receptors on the protein and the site of binding of an anion-transport inhibitor have also been studied.

Studies with inhibitors of erythrocyte anion transport have shown that the transport process is mediated by the major integral protein of the membrane (Cabantchik & Rothstein, 1974a; Lepke et al., 1976; Ship et al., 1977; Cabantchik et al., 1978). This protein, described as band 3 (Steck, 1974) or polypeptide 3 (Jenkins & Tanner, 1977a), has an apparent subunit mol.wt. of 90000-100000, and spans the erythrocyte membrane (Bretscher, 1971; Boxer et al., 1974). Studies on the protein have shown that it is glycosylated (Ho & Guidotti, 1975; Yu & Steck, 1975a; Tanner et al., 1976) and contains receptors for concanavalin A and the lectins from Phaseolus vulgaris and Ricinus communis (Findlay, 1974; Tanner & Anstee, 1976; Jenkins & Tanner, 1977b), which are located in the C-terminal region. The folding and disposition of the polypeptide chain of the protein into the membrane has been studied by chemical and enzymic cleavage. We have used the sites on the protein that are 'l25-labelled by lactoperoxidase in appropriate membrane preparations as markers for the cytoplasmic and extracellular location of regions of the protein (Boxer et al., 1974). The content of these sites in the major fragments obtained by trypsin cleavage of the protein at an intracellular region led us to suggest Abbreviation used: DIDS, 4,4'-di-isothiocyanatostilbene-2,2'-disulphonic acid. Vol. 181

that the polypeptide chain traversed the membrane at least twice (Jenkins & Tanner, 1975, 1977a). Steck et al. (1976, 1978) have also studied the proteolytic fragmentation of this protein and have shown that a large soluble fragment that binds the glycolytic enzyme glyceraldehyde 3-phosphate dehydrogenase (Yu & Steck, 1975b) can be obtained from a cytoplasmic domain of the protein. In fragmentation studies on the isolated protein, Drickamer (1976, 1977) used additional markers for the extra- and intra-cellular segments of the aniontransport protein and concluded that the polypeptide chain may cross the membrane four times. We have extended our studies on the topography of the anion-transport protein with further investigation of the major and minor products of proteolytic fragmentation at a known surface of the membrane. The order of these fragments in the polypeptide sequence, their origin and the topographical location of the lactoperoxidase-labelled sites contained in them suggest that the polypeptide chain contains at least four membrane-spanning segments and an additional loop that either also spans the membrane or is simply embedded in it. The general features of this model for the structure of the protein are similar to those suggested by Steck et al. (1978) and Grinstein et al. (1978) and differ from the model that we suggested previously (Jenkins & Tanner, 1975, 1977a) in not requiring the presence of a duplicated

478

D. G. WILLIAMS, R. E. JENKINS AND M. J. A. TANNER

set of sites that are labelled by lactoperoxidase and in the location of the N-terminus of the protein.

Methods Thermolysin or trypsin digestion of intact erythrocytes, labelling of erythrocytes or erythrocyte 'ghosts' by using Na'25I and lactoperoxidase, the isolation of labelled peptides and the preparation of thermolysin peptide 'maps' were done as described previously (Boxer et al., 1974; Jenkins & Tanner, 1975). After proteinase digestion of the erythrocytes, the cells were washed three times with 0.2 % bovine serumn albumin in 0.15M-NaCl, three times with 0.15M-NaCl and once with iso-osmotic sodium phosphate buffer, pH7.4 or 8.0, as indicated, before lysis. Lysis at pH 7.4 was done into cold hypo-osmotic sodium phosphate buffer, pH 7.4 (iso-osmotic sodium phosphate buffer, pH7.4, diluted with 19 vol. of deionized water) as described by Dodge et al. (1963). 'Ghosts' were depleted of bands 1, 2, 5 and 6, and digested with trypsin at low ionic strength as described by Steck et al. (1976). Trypsin treatment was terminated by adding di-isopropyl phosphorofluoridate (2,u1/mg of trypsin) and/or soya-bean trypsin inhibitor (1 mg/mg of trypsin). The preparation and labelling of erythrocytes with DIDS was carried out as described by Cabantchik & Rothstein (1974a) by using 94uM-DIDS. Sodium dodecyl sulphate/polyacrylamide-gel electrophoresis Electrophoresis in the presence of 0.1 % sodium dodecyl sulphate was done in slab gels with the buffers described by Fairbanks et al. (1971) or Laemmli (1970) as indicated. The separating gels contained a linear gradient of acrylamide concentration and were prepared by using a gradient maker. The ratio of acrylamide monomer to NN'-methylenebis(acrylamide) was maintained at 1: 30 throughout the gel. Protein stained gels containing a high concentration of acrylamide tended to crack on drying. This cracking was decreased by soaking the gel for about 3 h in a solution containing 40O% (v/v) methanol and 5 % (v/v) glycerol before drying between two sheets of water-permeable Cellophane (a gift from British Cellophane) on a porous polyethylene plate under vacuum. After this treatment the dried gels tended to be hygroscopic and were best stored in a desiccator. For the detection and elution of labelled fragments the unstained wet gels were covered with wrapping film and radioautographed when exposures of less than 8 h were used. The gel overlay was first removed to avoid artefactual blackening of the film because of the reducing agent in the samples. For longer

exposures the unstained gel was dried before radioautography. The fluorescence of DIDS-labelled bands in the gels was visualized under u.v. illumination after soaking the unstained gel in 40 % (v/v) methanol for 2 h. The fluorescent bands were marked by puncturing the gel with a needle and gels were subsequently stained for protein. The marker protein mixture used for calibrating apparent molecular weights of the fragments was described by Jenkins & Tanner (1977b), except that myoglobin (mol.wt. 17000) replaced cytochrome c. A commercial molecular-weight-marker peptide mixture (BDH) covering the mol.wt. range 170002500 was also used for low-molecular-weight fragments. Carboxypeptidase A digestion of erythrocytes Packed washed 125I-labelled erythrocytes or thermolysin-treated erythrocytes (0.2 ml in each case) were washed with 0.15 M-NaCl containing 0.1 MNH4HCO3. A 25% (v/v) suspension of the packed cells in the same solution was treated with 0.15mg of carboxypeptidase A (Worthington) at 37°C for 10min. The cells were washed three times with 0. 1 5 M-NaCl containing 0.2 % bovine serum albumin and three times with iso-osmotic sodium phosphate buffer, pH 7.4, before lysis.

Results and Discussion Previous studies have shown that when the human erythrocyte anion-transport protein was 125I-labelled in leaky erythrocyte membranes by using lactoperoxidase, isolated and digested with thermolysin, 14 major radioactive peptides were obtained that could be distinguished by their mobility on peptide 'maps'. Five of these peptides (numbered 1 to 5; see Figs. 2a and 4a) were shown to be derived from sites located in the extracellular region of the protein, whereas the remaining nine peptides (numbered 6 to 14; see Figs. 2a and Sa) were from sites in the cytoplasmic domain of the anion-transport protein (Boxer et al., 1974; Jenkins & Tanner, 1975, 1977a). In the present paper we have studied the fragmentation of the protein by the action of proteinases at a known side of the membrane. By using membranes that have been 125I-labelled by lactoperoxidase, and determining which of the radioactive peptides are present in thermolysin digests of the labelled fragments, it is possible to deduce both the topographical location of the fragments in the membrane and their location in the amino acid sequence of the polypeptide chain. The presence of the labelled peptides characteristic of the anion-transport protein in the peptide 'maps' of the thermolysin digest of the fragments also provides unambiguous evidence that they are derived from this protein, since thermo1979

STRUCTURE OF ERYTHROCYTE MEMBRANE ANION-TRANSPORT PROTEIN lysin digestion of the remaining major intrinsic membrane protein, the major sialoglycoprotein (which is also strongly 'l25-labelled by lactoperoxidase), gives rise to radioactive peptides that differ in mobility on peptide 'maps' from those of the anion-transport protein. However, to simplify the identification and purification of fragments from the anion-transport protein the '251-labelled membranes were 'depleted' or 'stripped' of extrinsic membrane proteins or soluble fragments, as appropriate, by using the techniques described by Steck et al. (1976). '251-labelled erythrocytes were also trypsin-digested to remove the radioactivity associated with the major erythrocyte sialoglycoprotein.

Soluble tryptic fragments of the anion-transport protein Although the extracellular region of the aniontransport protein is resistant to the action of trypsin under iso-osmotic ionic conditions, the cytoplasmic regions of the protein are susceptible to this proteinase under these conditions. Digestion of leaky erythrocyte 'ghosts' with trypsin at iso-osmotic ionic strengths yields a major membrane-bound fragment (TI; Jenkins & Tanner, 1975, 1977a). Steck et al. (1976) have also shown that fragments of the protein

T2s2-

x -T2*ie ':.F.:.'' ......S'.--T x-

-T2s

T4s2

T3sT3s-

;

s

.:.:

I-T3s -T4s

T5s-

T6s-

(a)

(b)

(c)

(d)

(e)

(f)

(gC

Fig. 1. Sodium dodecyl sulphate/polyacrylamide-gel electrophoresis of fragments solubilized from anion-transport protein by trypsin Erythrocyte 'ghosts' were depleted of peripheral proteins and the anion-transport protein crosslinked by incubation at 0°C for 30min with 100,pMCuS04, and 20OM-o-phenanthroline in 5mM-sodium phosphate, pH 8.0, as described by Steck et al. (1976). The 'ghosts' were digested with trypsin as indicated and digestion was terminated by adding 1p1 of diisopropyl phosphorofluoridate and further incubation at 0°C for 10min. After centrifugation at 40000g for Vol. 181

479

are also solubilized from the protein during trypsin digestion of leaky erythrocyte membranes at low

ionic strengths. We have studied the composition and origin of the soluble tryptic fragments released from the anion-transport protein on digestion of membranes at iso-osmotic and at low ionic strengths. Erythrocyte 'ghosts' were oxidized with the Cu2+/ phenanthroline reagent to form a disulphide-linked dimer of the anion-transport protein and depleted of peripheral proteins as described by Steck et al. (1976). The oxidized membranes were digested with trypsin at low ionic strength as described by Steck et al. (1976), and the fragments released from the membrane were separated by gel electrophoresis in

20min, the supernatant was mixed with 0.25 vol. of 5% sodium dodecyl sulphate, 50mM-Tris/HCI buffer, pH8.0, 5mM-EDTA, pH8.0, 20% glycerol and 2 mM-phenylmethanesulphonyl fluoride and heated at 100°C for 2min. The reduced samples were made 40mM with respect to dithiothreitol and heated at 100°C for 2min before electrophoresis. Gels (a)-(e) show the results of trypsin digestion by incubation for lh at 0°C with 0.5pg of trypsin/ml in 5mMsodium phosphate buffer, pH7.0. (a) and (b) show dried Coomassie Blue-stained gels of unreduced (a) and reduced (b) trypsin-solubilized material. (c) shows a radioautograph of a dried gel of a reduced sample from trypsin-solubilized material derived from erythrocyte 'ghosts' '251-labelled by using lactoperoxidase. Electrophoresis was on the same slab gel as that showii in (a) and (b). (d) and (e) show radioautographs of dried gels of unreduced (d) and reduced (e) samples of trypsin-solubilized material from erythrocyte 'ghosts' l25l-labelled by using lactoperoxidase. The samples were derived from a different experiment from that shown in (c). (f) and (g) show the results of trypsin digestion by incubation at 37°C for 15 min with 0.5 mg of trypsin/ml. Digestion at iso-osmotic ionic strength was carried out in 5 mM-sodium phosphate, pH 8.0, containing 0. 1 5 MNaCl, whereas digestion at low ionic strength was carried out in 5 mM-sodium phosphate, pH 8.0, alone. Both samples were derived from the $ame preparation of erythrocyte 'ghosts' that had been 1251-labelled by using lactoperoxidase. (f) shows a radioautograph of material solubilized by trypsin at iso-osmotic ionic strength. (g) shows a radioautograph of material solubilized by trypsin at hypoosmotic ionic strength. Sodium dodecyl sulphate/ polyacrylamide-gel electrophoresis was done by using slab gels containing a gradient of acrylamide concentration with the discontinuous buffer system of Laemmli (1970). The stacking gels all contained 5 % acrylamide and the separating gels contained acrylamide (w/v) concentration gradients of 8-25 % amide (sample a-c), 10-25% acrylamide (samples d and e), and 5-15 % acrylamide (samples f and g). The position of the various fragments is indicated on the gels.

480


the presence or absence of reducing agent. Fig. 1(b) shows that three major fragments of the anion-transport protein were obtained (T2s, T3s and T4s, with apparent mol.wts. of 40000, 23 000 and 20000 respectively) on electrophoresis in the presence of reducing agent. The additional band X of apparent mol.wt. 43000 was unlikely to be derived from the aniontransport protein since it was also obtained when no trypsin was added to the 'ghosts'. When no reducing agent was present in the electrophoresis samples (Fig. la) fragments T2s and T4s were obtained in much lower yield and new bands of apparent mol.wt. 80000 and 38000 were obtained, which represent disulphide-linked dimers of fragments T2s and T4s respectively. Further new bands of apparent mol.wt. 58000 and 17000 were also obtained, which are probably disulphide-linked complexes of fragments T2s and T4s, and of low-molecular-weight fragments respectively. These results are in agreement with those of Steck et al. (1976). The same experiment was carried out with intact erythrocytes that had been '251-labelled with lactoperoxidase and were then trypsin-digested to remove the radioactivity associated with the major sialoglycoprotein. Trypsin treatment of membranes derived from these cells did not release any radioactivity in the soluble peptides. When the experiment was performed with leaky erythrocyte 'ghosts' that had been 1251-labelled with lactoperoxidase, electrophoresis in the presence of reducing agent showed that fragments T2s and T3s contained radioactive label (Figs. Ic and le), whereas no label was found in fragment T4s. Lower-molecular-weight radioactive fragments were also present. When electrophoresis was done in the absence of reducing agent fragment T2s was obtained in much decreased yield and its disulphide-linked dimer (fragment T2s2)

appeared (Figs. Id and le). The same fragments were obtained if digestion was done under the same conditions, but at iso-osmotic ionic strength rather than at hypo-osmotic ionic strength. When the temperature of digestion and the trypsin concentration were increased more extensive digestion occurred. Figs. l(f) and l(g) show the results obtained by using 1251-labelled erythrocyte 'ghosts' and digestion with 0.5mg of trypsin/ml of 'ghosts' at 37°C for 15min at hypo-osmotic and iso-osmotic ionic strength. No fragment T2s was obtained and the yield of fragment T3s was low. Digestion at iso-osmotic ionic strength produced further major radioactive fragments T5s, Y, and T6s, as well as other minor fragments (Fig. If). Fragments Y and T6s migrated with the Bromophenol Blue dye front on discontinuous sodium dodecyl sulphate/polyacrylamide-gel electrophoresis on gradients of acrylamide up to 25 % (w/v), and both these and fragment T5s migrated in front of the lowest-molecular-weight marker peptide (mol.wt.

2512), which was present in the marker-peptide mixture used for molecular-weight calibration. Fragment T5s yielded an extrapolated apparent mol.wt. of 1000. Although no estimates of the molecular weights of fragments Y and T6s could be made, they were clearly very small fragments. Sodium dodecyl sulphate/polyacrylamide-gel electrophoresis on gradients of acrylamide concentration (8-15 % acrylamide) with the buffers described by Fairbanks et al. (1971) gave the markedly different value for the apparent mol.wt. of fragment T5s of 15000, but fragments Y and T6s migrated as a single band with the Bromophenol Blue dye front in this system. When the gels were stained with the Coomassie Blue stain no radioactivity corresponding to fragments T6s and Y was detected and the amount of radioactivity corresponding to fragment T5s was markedly decreased. These fragments do not remain fixed in the gel after the protein-staining procedure. The radioactive fragments from the aniontransport protein were isolated from leaky erythrocyte membranes that had been "251-labelled with lactoperoxidase and digested with trypsin by using the conditions of Figs. 1(c) and 1(f). The isolated fragments were digested with thermolysin and the peptides separated on peptide 'maps' (Fig. 2). Fragments T2s, T3s and T5s all gave the same peptide 'map'. These contained only the labelled peptides 6, 7, 11, 12 and 13, all of which are derived from sites located in the cytoplasmic region of the protein, and some minor radioactive peptides that are also derived from the anion-transport protein (Figs. 2b and 2c). Fragment T6s (Fig. 2d) gave a peptide 'map' that contained only the labelled peptides, 8, 9 and 10, which are also derived from sites that are located in an intracellular domain of the protein. The peptide 'map' of fragment Y did not contain any of the labelled peptides characteristic of the anion-transport protein and is probably not derived from this component. It is evident that fragment T2s corresponds to the soluble 41 000mol.wt. tryptic fragment reported by Steck et al. (1976, 1978), whereas fragments T3s and T4s correspond to the mol.wt. 22000 and 16000 fragments also described by these workers. The cause of the discrepancy in apparent molecular weight that we obtained for fragment T4s (20000), and the value of 16000 obtained by Steck and his co-workers is not clear, but it may be related to the different acrylamide-gel system that we have used. Steck et al. (1978) suggested that fragments T3s and T4s are subfragments of fragment T2s. Our results confirm that this is indeed the case since fragments T2s and T3s share the same set of "251-labelled sites and fragments T2s and T4s both contain the cysteine residue involved in the oxidative dimerization of the anion-transport protein. The very small fragment T5s is the product of extensive digestion of fragment

1979

481

STRUCTURE OF ERYTHROCYTE MEMBRANE ANION-TRANSPORT PROTEIN T3s and contains all the 125I-labelled sites present in fragment T3s. It is clear that these sites (peptides 6, 7, 11, 12 and 13) must be clustered in a small region of fragment T3s. Steck et al. (1978) concluded that the fragment corresponding to T2s contains the blocked N-terminus of the protein and represents a major cytoplasmic domain of the anion-transport protein. They also showed that the fragment corresponding to T3s contains the N-terminus of the intact protein and of fragment T2s, whereas the fragment corresponding to T4s is derived from the C-terminal region of fragment T2s (see Fig. 8a). This alignment places the major cytoplasmically located labelled sites 6, 7, 11, 12, 13 within 23000 daltons from the N-terminus of the protein and more than 1600020000 daltons from the point where the polypeptide enters the membrane. The cytoplasmically located sites 8, 9, 10 are also clustered together in the very small fragment T6s, but these are located in different cytoplasmic regions not present in the large fragment T2s.

Major and minor membrane-bound fragments of the anion-transport protein from thermolysin digestion of intact erythrocytes Thermolysin digestion of intact erythrocytes yields a major membrane-bound fragment of the anion-transport protein of approx. mol.wt. 60000 (fragment 3f; Jenkins & Tanner, 1975). Fragments of indistinguishable mobility on sodium dodecyl sulphate/polyacrylamide-gel electrophoresis are also produced by a variety of other proteinases including chymotrypsin (fragment Chy A; Drickamer, 1976; Steck et al., 1976), subtilisin and Pronase (Bender et al., 1971; Jenkins & Tanner, 1975). There have been several reports in the literature that treatment of intact erythrocytes by this group of enzymes also produces a fragment from the anion-transport protein of apparent mol.wt. approx. 40000 (Cabantchik & Rothstein, 1974b; Jenkins & Tanner, 1975; Steck et al., 1976). This fragment migrated as a diffuse band on sodium dodecyl sulphate/polyacrylamide-gel electrophoresis and was obtained in variable and low yields. Higher yields of the fragment

(b) (a)

r 4 5

3.

0.11

6.-07 8

g

1

(d)

(c) eg

1

o 8

13# ..

9

W

r:11 :.. ...

.:

I

Fig. 2. Thermolysin peptide 'maps' of soluble trypsin fragments ofanion-transport protein Radioautographs of soluble tryptic fragments derived from erythrocyte 'ghosts' '25I-labelled with lactoperoxidase. Electrophoresis was done so that labelled peptide 14 migrated off the 'maps' (compare with Fig. 5a) so as to obtain better resolution of the region around peptides 6, 11, 12 and 13. (a) Intact protein. (b) Fragment T3s is isolated from the sample shown in Fig. l(c). (c) Fragment T5s isolated from the sample shown in Fig. 1(f). (d) Fragment T6s isolated from the sample shown in Fig. 1(g). Abbreviations: C, chromatography dimension; E, electrophoresis (pH 3.5) dimension. Q Vol. 181

482


Washed erythrocytes 1251-labelling with lactoperoxidase

Labelled erythrocytes Thermolysin digestion

Thermolysin-digested labelled erythrocytes Wash with (i) 0.2 % bovine serum albumin in 0.15 M-NaCI and (ii) iso-osmotic sodium phosphate buffer, pH 7.4

Washed thermolysin-digested labelled erythrocytes Carboxypeptidase A digestion; wash; Hypo-osmotic lysis; wash

Hypo-osmotic lysis; wash

'Ghosts'

'Ghosts' 1251-labelling with lactoperoxidase; wash

Membranes from thermolysin-treated labelled erythrocytes (preparation A) Isolation of fragments ThiThS by gel electrophoresis (see Figs.3a-3d for results for preparation A)

Thermolysin digestion; preparation of peptide 'maps'

'Strip' with 0.1 M-NaOH; wash with hypo-osmotic

Strip' with 0.1 M-NaOH; wash with hypo-osmotic 5,;odium phosphate buffer, ppH 7.4

'Strip' with 0. 1 M-NaOH; wash with hypo-osmotic M sodium phosphate buffer, pH7.4

sodium phosphate buffer,

pH 7.4

Labelled membranes from thermolysin-treated labelled erythrocytes (preparation B)

Isol4ation of fragments Th1-

Th55 by gel electrophoresis (se e Figs.3a-3d for results for preparation A)


Membranes from carboxypeptidase-treated labelled erythrocytes (preparation C) Isolation of fragments ThITh5 by gel electrophoresis (see Figs.3a-3d for results for preparation A)


Peptide 'maps' of fragments Radioautographs of Radioautographs of peptide Thl-Th5 from preparation C "maps' of fragments Thl-Th5 peptide 'maps' of (Table 3) from preparation A (Fig. 4) preparation B (Fig. 5) Scheme 1. Isolation offragments Thl-Th5 were obtained when enzyme treatment was terminated by an inhibitor and when the treated cells were washed with serum albumin solutions (Cabantchik

& Rothstein, 1974b). The fragmentation products from thermolysin cleavage of the anion-transport protein were studied by thermolysin digestion of lactoperoxidase-labelled erythrocytes followed by serum albumin washing of the cells (preparation A, Scheme 1). Pretreatment of the erythrocytes with trypsin was not used because a small amount of digestion of the anion-transport protein occurred under these conditions yielding traces of fragments that interfered with the sub-

sequent analysis of the thermolysin-digestion products. Treatment of erythrocytes with thermolysin alone (Figs. 3b and 3c) caused partial digestion of the major sialoglycoprotein and fragments derived from the sialoglycoprotein were present in these digests, but these could be distinguished by the characteristic mobility on thermolysin peptide 'maps' of labelled peptides from the sialoglycoprotein. In addition to the major fragment Th 1 (the fragment referred to as 3f or Chy A above), a number of minor bands were obtained in 'ghosts' from thermolysin-treated erythrocytes (fragments Th2-Th5; Figs. 3a-3d). The fragment Th2 region was an ill-

1979

STRUCTURE OF ERYTHROCYTE MEMBRANE ANION-TRANSPORT PROTEIN

3--

-~~S as

Thl-

-Th l Th2

Th1Th2{ Th3..T...5

:::..:

'O,

Th3Th4 Th5- i Th4Th 5-

Th6-

:.L

(a) (b)

(c)

(d)

(e) (f)

Fig. 3. Sodium dodecyl sulphate/polyacrylamide-gel electrophoresis of membrane-bound fragments obtained after extracellular thernmolysin cleavage of the aniontransport protein Radioautographs of portions of dried slab gels. (a) shows the membrane-bound fragment from thermolysin-treated '251-labelled erythrocytes (preparation A, Scheme 1). The control sample (b) from the same preparation was treated in the same way, but thermolysin digestion was omitted. Separation was on a gel containing a gradient of 5-15 % (w/v) acrylamide with a 4% acrylamide overlay by using the buffers described by Fairbanks et al. (1971). (c) and (d) show samples from an experiment by the same procedure as the samples in (a) and (b) separated on a gel containing a higher concentration of acrylamide. Separation of the control (c) and membranes from thermolysin-digested 1251-labelled erythrocytes (d) was on a gel containing a gradient of 8-25 % (w/v) acrylamide and a 4% acrylamide overlay, with the buffers described by Fairbanks et al. (1971). No fragment Th5 was detected in this experiment and the small amounts of fragment Th4 obtained were detected on more extended exposure of the radioautographs. (e) and (f) show membranebound fragments obtained on thermolysin digestion of resealed 125I-labelled erythrocyte 'ghosts'. Erythrocytes were washed with iso-osmotic sodium phosphate buffer, pH 6.0, and were lysed in cold iso-osmotic sodium phosphate buffer, pH 6.0, which had been diluted with 19.5 vol. of deionized water. The 'ghosts' were washed with hypo-osmotic phosphate buffer, pH7.4, 125I-labelled by using lactoperoxidase, and washed three times with cold hypo-osmotic phosphate buffer, pH7.4. The 'ghosts' were resealed by adding 20 vol. of 0.15M-NaCi containing 4mM-MgSO4 and incubated at 37°C for 45 min. After centrifugation at 40000g for 20min, the 'ghosts' were suspended in 2 vol. of 0.1SM-NaCI containing 4mM-MgSO4 and layered on a sucrose 'cushion' containing 43 % sucrose, 25mM-NaCl and 25mM-Tris/HCI, pH7.5, and centrifuged at 65000gav. for 1h in a swing-out

Vol. 181

483

defined area migrating in front of and up to the fragment Thl band. Fragment Th3 migrated as a broad band and was more clearly defined and better separated on gels containing a high acrylamide concentration (Fig. 3d). The same labelled bands were found in six separate experiments. The yields of the fragments were somewhat variable and fragment Th5 was only obtained in four of these experiments. Fragments Thl-Th5 were isolated and redigested with thermolysin and the radioactive peptides separated on peptide 'maps' (preparation A, Scheme 1). The peptide 'map' of fragment Thl (Fig. 4b) contained peptides 1, 2 and 3 and smaller amounts of peptides 4 and 5 than the labelled intact protein from which it was derived (Fig. 4a). Fragment Th3 contained peptides 4 and 5 only (Fig. 4c), whereas fragments Th4 and Th5 gave similar peptide 'maps', which contained peptides 1, 2 and 3 with traces of peptide 4 (Fig. 4d). In the above experiment the protein was 1251_ labelled only in the extracellular region. To label the protein fragments in both intracellular and extracellular domains, membranes from thermolysindigested erythrocytes that had been 1251-labelled with lactoperoxidase were again '251-labelled with lactoperoxidase and the labelled peripheral proteins removed by 'stripping' with 0.1 M-NaOH (preparation B, Scheme 1). No new fragments were obtained compared with digests from erythrocytes that had been labelled at the extracellular surface only. The thermolysin peptide 'map' of fragment Thl (Fig. 5a) from this experiment was similar to that of the intact protein (Fig. 2a), except that only traces of peptides 8, 9 and 10 were observed. Fragment Th3 contained peptides 4 and 5 and peptides 8, 9 and 10 (Fig. Sb). Fragments Th4 and Th5 contained peptide 14 in addition to peptides 1, 2 and 3 (Fig. Sc). Band Th2 appeared to be impure. Band Th2 contained a fragment rich in peptides 4 and 5 and peptides 8, 9 and 10. rotor. The resealed labelled 'ghosts' were washed with 0.15 M-NaCl and assayed for sealing by measuring the activity of glyceraldehyde 3-phosphate dehydrogenase in the absence and presence of Triton X- 100 as described by Boxer et al. (1974). Thermolysin digestion of the resealed labelled 'ghosts' and subsequent steps were done as shown in Scheme 1 (preparation A) for labelled erythrocytes. (e) shows membranes from thermolysin-digested resealed 1251labelled 'ghosts'. (f) is a control from the same preparation as that shown in (e), but with the thermolysin-digestion step omitted. Separation was on a gel containing a gradient of 8-15% (w/v) acrylamide and a 4% acrylamide overlay by using the buffers described by Fairbanks et al. (1971). Abbreviations: 3, intact anion-transport protein; S,

major sialoglycoprotein.

wm_,~ ~ ,~ . . :.{'t


484 (a)

(b)

4 5

24~

~

~

2

3** 1

*f :':.S+ .:.A

-1

E

+

Fig. 4. Peptide 'maps' of thermolysin fragments of anion-transport protein 1251-labelled in the extracellular domain Radioautographs are shown of peptide 'maps' of thermolysin digests of fragments purified by sodium dodecyl sulphate/ polyacrylamide-gel electrophoresis of samples derived as shown in Scheme 1 (preparation A). (a) Intact anion-transport protein from a separation similar to that shown in Fig. 3(b). (b) Fragment Thl from a separation similar to that shown in Fig. 3(a). (c) Fragment Th3 from a separation similar to that shown in Fig. 3(d). (d) Fragment Th4 from a separation similar to that shown in Fig. 3(a). Abbreviations: C, chromatography dimension; E, electrophoresis (pH 3.5) dimension.

We reported previously that fragment ThI contained the same set of lactoperoxidase-labelled sites as the intact anion-transport protein (Jenkins & Tanner, 1975, 1977a). In these previous experiments fragment Thl was purified by electrophoresis in gels containing 8 % acrylamide and these give much inferior resolution compared with the acrylamidegradient gels used in the present study. In our present experiments, the labelled peptides 8, 9 and 10 were hardly detectable in fragment Thl (compare Figs. 2a and 5a), whereas peptides 1, 4 and 5 were present in lower amounts than in the whole protein (compare Figs. 4a and 4b). The Thl preparations in Figs. 5(a) and 4(b) were derived from the labelled intact proteins shown in Figs. 2(a) and 4(a) respectively. This suggested that peptides 4 and 5 and peptides 8, 9 and 10 observed previously in peptide 'maps' of fragment Th I might be due to contaminating material, perhaps from the Th2 band, which was not resolved during electrophoresis. The radioactivity in the labelled peptides 1-5 found in the thermolysin peptide 'maps' of the intact protein that had been 125I-labelled in the extracellular region was measured and compared with the radioactivity in these peptides on peptide 'maps'

from fragments Thl, Th2 and Th4 derived from the same labelled-protein preparation (Table 1). Fragments Thl and Th4 were very similar and gave substantial recovery of the labelled peptides 1, 2 and 3, but low recoveries of peptides 4 and 5. A substantial proportion of the peptide 5 in fragment Thi is probably derived from contamination by the adjacent band Th2, which is very rich in this peptide. Although fragments Thl and Th4 gave only about 30% of the amount of labelled peptide 1 obtained from the intact protein, the similar and substantial recovery of this peptide in fragments Thl and Th4 and fragment ThT (discussed below) suggests that it really is derived from these fragments. Unpublished work (M. J. A. Tanner, D. G. Williams & D. Kyle) suggests that the labelled peptides 1 and 3 are related and may contain the same labelled tyrosine residue(s) of the intact protein. The relative proportions of peptides 1 and 3 obtained in peptide 'maps' is influenced by prior cleavage of the protein near site 3, and site 3 is close to the C-terminus of fragments Thl and Th4 (see below). These results suggest that peptides 4 and 5 in the peptide 'maps' of fragment Thl do result from contamination of material in the Th2 band.

1979


485

(b)

(a)

I....

....

.1 4 45

* 14 .* 07

'jo

5 WI

*..

:.

* 2-

..U

g

124

94 1 w....

13 .I

C

2E

-

.-

E

4

Fig. 5. Peptide 'maps' of thermolysin fragments of anion-transport protein 125I-labelled in both the extracellular and intracellular domains Radioautographs of peptide 'maps' of thermolysin digests of fragments purified by sodium dodecyl sulphate/polyacrylamide-gel electrophoresis. The peptide 'maps' (a-c) were derived as shown in Scheme 1 (preparation B). (a) Fragment Thl, from a separation similar to that shown in Fig. 3(a). (b) Fragment Th3, from a separation similar to that shown in Fig. 3(d). (c) Fragment Th4 from a separation similar to that shown in Fig. 3(c). (d) Fragment Th6 from thermolysin-digested resealed 1251-labelled 'ghosts'. The sample, from a separation similar to that shown in Fig. 3(f), was derived as described in the legend to Fig. 3(f). Fig. 2(a) shows the peptide 'map' of the intact anion-transport protein from the same preparation as that shown in (a). Abbreviations: C, chromatography dimension; E, electrophoresis (pH 3.5) dimension.

In the experiment using preparation B (Scheme 1), fragmentation was done on the intact erythrocytes and labelling was subsequently done on the 'ghosts' containing fragment Thl prepared from the digested cells. However, the control (Fig. 2a) was prepared by labelling the intact protein in erythrocyte 'ghosts'. It was conceivable that fragment Thl contained the amino acid sequences corresponding to sites 8, 9 and 10, but that the lack of labelling in sites 8, 9 and 10 of fragment Thl (Fig. 5a) was due to a decreased susceptibility to labelling of fragment Thl compared with the intact protein at these sites. To clarify this point, the intact protein was labelled at both intracellular and extracellular domains before fragmentation. In this experiment leaky erythrocyte 'ghosts' were l25l-labelled with lactoperoxidase and resealed as described by Passow (1969). The resealed labelled 'ghosts' were purified as described by Bodeman & Passow (1972), and subsequently digested with thermolysin. No activity of glyceraldehyde 3-phosphate dehydrogenase could Vol. 181

be detected in the resealed-'ghost' preparation unless Triton X-100 was present in the assay medium, confirming that the membranes were sealed and had the same orientation as the intact erythrocyte. Fig. 3(e) shows a radioautograph of the sodium dodecyl sulphate/polyacrylamide-gel electrophoretogram of the fragments obtained. In addition to fragments Thl-Th5, a low-molecular-weight fragment Th6, of apparent mol.wt. 16000, was resolved on this gel. The radioactivity in peptides 8, 9, 10 and 11 was measured in the peptide 'maps' of fragment Thl from this experiment and the intact labelled protein from which it was derived. When normalized to equal amounts of radioactivity in peptide 11, peptides 8, 9 and 10 of fragment Thl contained 15, 14 and 10 % respectively of the radioactivity in the same sites from the intact protein. The peptide 'map' of fragment Th2 obtained from this experiment was again found to be rich in sites 8, 9 and 10, suggesting that the presence of small amounts of these labelled peptides in the ThI preparation was due to contamina-


486

tion by fragment Th2. These results further confirm that fragment Thl does not contain peptides 8, 9 and 10. The peptide 'maps' of fragments Th3, Th4, Th5 contained the same labelled peptides as observed in the fragments from preparation B (Scheme 1) and shown in Figs. 5(a)-5(c). Peptide 'maps' of fragment Th6 (Fig. 5d) contained only peptides 4 and 5. Table 2 summarizes the labelled peptides found in peptide 'maps' of the fragments derived from extracellular thermolysin cleavage of the anion-transport protein.

Carboxypeptidase A treatment of the fragments Treatment of 125I-labelled undigested erythrocytes with carboxypeptidase A did not cause any change in the recovery of peptides 1-5 in the peptide 'maps' of the anion-transport protein. Carboxypeptidase A treatment of thermolysin-digested labelled erythrocytes (preparation C, Scheme 1) did not change the mobilities on sodium dodecyl sulphate/polyacrylamide-gel electrophoresis of the labelled fragments. Peptide 'maps' were done on corresponding fragments derived from thermolysin-digested 1251_ labelled cells before and after carboxypeptidase A treatment. Table 3 shows the relative radioactivity

Table 1. The recovery of extracellularly labelled peptides in thermolysin fragments relative to the intact protein The fragments Thl, Th2 and Th4 were obtained by using the procedure for preparation A (Scheme 1). The intact protein was isolated as described by Jenkins & Tanner (1977a) except that electrophoresis was done on a gel containing a concentration gradient of acrylamide. All the samples were derived from the same batch of '25I-labelled erythrocytes. The radioactivity in each of the labelled peptides in peptide 'maps' of the samples was estimated as described by Jenkins & Tanner (1975). For each peptide 'map' the radioactivity in the peptides was normalized so that the radioactivity in peptide 2 was 1.0. The recovery is the normalized radioactivity of a given labelled peptide in a particular fragment divided by the normalized radioactivity in the same labelled peptide in the peptide 'map' of the intact protein. Recovery of radioactivity

Peptide Fragment ... Thl I 0.32 2 1.0 3 1.17 4 0.046 5 0.18

Th2 0.28 1.0 1.07 0.50 3.06

Table 3. Effect of carboxypeptidase A treatment on the extracellular thermolysin fragments of the anion-transport protein The radioactivity in labelled peptides 1-5 was measured in peptide 'maps' of the correspoding fragments derived from thermolysin-treated erythrocytes and carboxypeptidase A-treated thermolysin-digested '251-labelled erythrocytes (Scheme 1, preparation C). The radioactivity of the peptides in each peptide map was normalized to equivalent amounts of peptide 2 for fragments Thl and Th4, and peptide 4 for fragment Th3. The normalized radioactivity in the labelled peptides from the carboxypeptidase-treated fragment was divided by the normalized radioactivity in the same peptide from the corresponding fragment to yield the recovery of the labelled peptide. indicates that the peptide was very weak or absent on the peptide 'map'. Recovery of peptide

Th4 0.24 1.0 1.17 0.043 0.014

Peptide Fragment number ... 1 Thl 2.0 Th3 Th4 2.3

2 1.0 -

1.0

3 0.12 0.09

4

5

1.0

1.0

-

Table 2. Apparent molecular weight and content of labelled peptides in peptide 'maps' of the membrane-bound thermolysin fragments of the anion-transport protein Peptides 1-5 are derived from extracellular sites and peptides 6-14 are from sites located in the cytoplasmic region of the protein. + indicates the presence of the labelled peptide in peptide 'maps' of the fragment, whereas - indicates its absence. Molecular weights were determined by sodium dodecyl sulphate/polyacrylamide-gel electrophoresis by using the buffers described by Fairbanks et al. (1971) on gels containing a gradient of 8-15% (w/v) acrylamide. Presence or absence of labelled peptide in peptide 'map' Labelled-peptide Fragment 10-4xApparent mol.wt. number ... 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Thl 60 -++ ---+++ Th3 42 Th4 24 + + + ..+ Th5 22.5 + + Th6 16 ThT 21.5 +

+

+-

+

+

+_

_

_

_

+

_

1979


487

Washed erythrocytes I

1251-labelling with lactoperoxidase; wash

Labelled erythrocytes (i) Thermolysin treatment (ii) Wash with 0.2% bovine serum albumin in 0.15 M-NaCI (iii) Wash with iso-osmotic sodium phosphate, pH7.4

Thermolysin-treated labelled erythrocytes I

Hypo-osmotic lysis

'Ghosts' from thermolysin-treated labelled erythrocytes (i) Depletion of bands 1, 2, S and 6 (ii) Digestion with trypsin (0.5mg/ml) in hypo-osmotic sodium phosphate buffer, pH7.4, containing 0. 15 M-NaCI at 370 C for 45 min (iii) Treatment with di-isopropyl phosphorofluoridate + 1 mg of soya-bean trypsin inhibitor/ml (iv) 'Strip' with 0.1 M-NaOH (v) Wash with hypo-osmotic sodium phosphate, pH 7.4

Trypsin-treated 'ghosts' from thermolysin-treated labelled erythrocytes I

Sodium dodecyl sulphate/polyacrylamide-gel electrophoresis (cf. Fig. 6b)

Fragment ThT Scheme 2. Preparation offragment ThT

q il! 3

-

2

X

of peptides 1-5 in the peptide 'maps' of fragments Thl, Th3 and Th4 from carboxypeptidase A-treated thermolysin-treated cells. The peptide 3 in fragments ThI and Th4 was almost totally removed by carboxypeptidase A treatment. There was also an increase in the amount of peptide 1 or a smaller decrease in the amounts of peptide 2 in these fragments. Peptides 4 and 5 of fragment Th3 were resistant to carboxy-

_I-ThT

(a)

(b)

Fig. 6. Trypsin

treatment

of 'ghosts' from thermolysin-

treated 1251-labelled erythrocytes and peptide 'map' of fragment ThT (a) and (b) show radioautographs of dried slab gels containing a gradient of 8-15% (w/v) acrylamide, and a 4% acrylamide overlay. Electrophoresis was done by using the buffers described by Fairbanks et

Vol. 181

al. (1971). The digest sample (b) was obtained as shown in Scheme 2. The control sample (a) was derived from the same preparation except that trypsin treatment was substituted for thermolysin treatment of the erythrocytes and the trypsin treatment of the 'ghosts' was omitted. Abbreviation: 3, intact anion-transport protein. (c) Radioautograph of peptide 'map' of thermolysin digest of fragment ThT '25l-labelled in both the extracellular and intracellular regions. The sample was derived as described in Scheme 2, except that after lysis of the thermolysin-treated labelled erythrocytes, the 'ghosts' were again 5251-labelled by using lactoperoxidase. Fragment ThT was isolated from a separation similar to that shown in (b).

488


peptidase A treatment. These results suggest that fragments ThI and Th4 share a common C-terminus, which yields peptide 3. This is consistent with our previous finding that the thermolysin cleavage that leads to fragment Thl occurs around the site that yields peptide 3 (Jenkins & Tanner, 1975) and the presence of tyrosine as the C-terminal residue of fragment Thl (Jenkins & Tanner, 1977b) and the comparable chymotrypsin fragment Chy-A (Drickamer, 1976). Trypsin digestion of 'ghosts' derivedfrom thermolysintreated erythrocytes Trypsin digestion at iso-osmotic ionic strengths of thermolysin-treated lactoperoxidase-labelled erythrocytes (Scheme 2) yielded a single major labelled membrane-bound fragment of the anion-transport protein (fragment ThT, Fig. 6b). This band was also the only major protein-staining band obtained under these conditions. A band of the same electrophoretic mobility was also obtained as the only major band when 'ghosts' from thermolysin-treated cells were digested with chymotrypsin at iso-osmotic ionic strengths. Peptide 'maps' of fragment ThT isolated after trypsin treatment of 'ghosts' from thermolysin-treated 125I-labelled cells, or after trypsin treatment of 125I-labelled 'ghosts' from thermolysin-treated cells contained the same labelled peptides (Fig. 6c). Thus even when the protein was labelled in both intra- and extra-cellular regions, only the extracellular peptides 1, 2 and 3 were obtained in fragment ThT. Fragment ThT had a slightly lower apparent molecular weight than fragment Th5 (Table 2) and gave a similar peptide 'map', except that labelled peptide 14 was absent from fragment ThT. Since the site yielding peptide 3 is located at the C-terminus of fragment Thl, fragment ThT must be derived from the C-terminal region of fragment Thl. The site giving rise to peptide 14 is therefore located in the N-terminal side of the trypsin cleavage leading to fragment ThT (Fig. 8a). Fragments similar to fragment ThT, but with an estimated mol.wt. of 17000, have been obtained by chymotrypsin treatment of erythrocytes followed by trypsin treatment of 'ghosts' (Grinstein et al., 1978) and by chymotrypsin treatment of erythrocyte 'ghosts' (Steck et al., 1976, 1978). These fragments may be of a similar size to fragment ThT, and the different value for the apparent molecular weight that we obtain may result from the different acrylamide-gel system that we have used (as is also found with fragment T4s).

Fragments that bind DIDS DIDS is a specific covalently bound inhibitor of erythrocyte anion transport that labels essentially only the anion-transport protein under conditions

that result in total abolition of erythrocyte anion transport (Cabantchik & Rothstein, 1974a; Passow et al., 1977; Ship et al., 1977). Erythrocytes were incubated with DIDS, washed with serum albumin solutions, and fragmentation was performed subsequently. The fragments that covalently bound DIDS were detected after gel electrophoresis by visualization of the fluoresence of bound DIDS in the unstained gel on illumination with u.v. light. DIDS fluorescence was found associated with fragments Thl and ThT and the major trypsinproduced fragment, TI (Jenkins & Tanner, 1977b). Although fragments Th4 and Th5 might be expected to bear DIDS fluorescence, none was detected in these bands, probably because of the low yields of these fragments that are obtained and the relative insensitivity of the detection method. These results confirm those of other workers who used radioactive DIDS or dihydro-DIDS on equivalent fragments (Cabantchik & Rothstein, 1974b; Lepke & Passow, 1976; Grinstein et al., 1978) and suggest that the binding site for DIDS is in the C-terminal region of fragment Thl. Lectin receptors on fragments of the anion-transport protein The anion-transport protein contains receptors for concanavalin A and the lectins from R. communis and P. vulgaris (Findlay, 1974; Tanner & Anstee, 1976). The oligosaccharides binding these lectins are present in the tryptic fragment TI, which contains the C-terminal region of the protein (Jenkins & Tanner, 1977b). Fig. 7 shows the pattern of binding of radioactive concanavalin A to 'ghosts' from thermolysin-treated erythrocytes, and to trypsintreated 'ghosts' from thermolysin-treated erythrocytes. The gels show some non-specific binding to the most intense protein-staining bands, but specific binding (which could be displaced by methyl amannoside) was obtained to a band corresponding to fragment Th3 in the 'ghosts' from thermolysintreated erythrocytes (Fig. 7a). In addition, a broad area of concanavalin A binding of lower mobility than fragment Thl was found in this sample. No corresponding protein-staining band was evident, and no binding in this region was found in 'ghosts' from untreated erythrocytes. When trypsin-treated ghosts from thermolysin-treated erythrocytes were examined (Fig. 7b) specific binding of concanavalin A in the region of fragment Th3 was again obtained, with some binding to fragment ThT that appeared to be non specific in nature, since it could not be displaced by methyl a-mannoside. Similar binding experiments with the lectin from R. communis gave results (not shown) that were less clear, since several membrane components other than the aniontransport protein bind this lectin. 'Ghosts' from thermolysin-treated erythrocytes showed binding in 1979


A

2

0

4

6

(b)

8

10

Th3

12

ThT

A

0

2

4

6

8

10

12

Distance along gel (cm)

Fig. 7. Binding of 12I5-labelled concanavalin A to fragments of the anion-transport protein (a) shows membranes from thermolysin-treated trypsin-digested erythrocytes. Trypsin-treated erythrocytes were digested with thermolysin and washed twice with 0.2 % bovine serum albumin in iso-osmotic sodium phosphate buffer, pH 7.4. The cells were washed four times with iso-osmotic sodium phosphate, pH7.4, and lysed into hypo-osmotic sodium phosphate buffer, pH 7.4. The 'ghosts' were washed in hypo-osmotic sodium phosphate buffer, pH7.4, until free of haemoglobin and depleted of bands 1, 2, 5 and 6, as described by Steck et al. (1976). (b) shows membranes from trypsin-treated 'ghosts' derived from thermolysin-treated trypsin-digested erythrocytes. Membranes prepared as described for (a) were digested with 1 mg of trypsin/ml at isoosmotic ionic strength. After quenching the reaction with di-isopropyl phosphorofluridate, the membranes were washed three times in hypo-osmotic sodium phosphate buffer, pH7.4. Each sample was run in duplicate on a sodium dodecyl sulphate/polyacrylamide-slab gel containing a gradient of 5-15 % (w/v) acrylamide with a 4% acrylamide overlay by using the buffers described by Fairbanks et al. (1971). Vol. 181

489

the region corresponding to fragment Th3. Trypsintreated 'ghosts' from thermolysin-treated cells showed similar binding in this region, but no binding to fragment ThT was obtained. These results suggest that fragment Th3 contains the receptors for concanavalin A and the R. communis lectin, and that these receptors are not present in fragment ThT. The latter observation is not unexpected since fragment ThT is derived from fragment Thl, which does not contain the lectin receptors (Jenkins & Tanner, 1977b). Neither in the present study, nor in previous ones, were any lectin receptors detected in the region corresponding to fragment Thl, although analyses of the isolated fragment showed the presence of carbohydrates in fragment ThI (Jenkins & Tanner, 1977b). The possibility remains that this carbohydrate reflected the presence of tightly bound contaminating glycolipids in the preparations, as suggested by Drickamer (1978). Alignment ofthe fragments on the amino acid sequence of the protein Fig. 8(a) shows the location of the fragments on the linear amino acid sequence of the protein and the partial order of the sites that are "25I-labelled with lactoperoxidase. It is assumed in Fig. 8(a), for the purpose of discussion, that each of the major radioactive peptides found in peptide 'maps' of the intact protein corresponds to a different '25l-labelled tyrosine site in the protein. This is not necessarily the case since, although exhaustive thermolysin digestion was used to prepare the peptides, a single labelled site could yield two or more overlapping peptides. The labelled sites in the protein are assigned the numbers of the corresponding labelled peptides. Fragments ThI and Th3 contain the sum of the The duplicate tracks were separated, fixed, and incubated with 125I-labelled concanavalin A (1mg containing 108 c.p.m. of 1251) as described by Tanner & Anstee (1976). One track was incubated in the absence of inhibitor, whereas the other track was incubated in the presence of 0.3 M-methyl amannoside, 0.3 M-methyl a-mannoside being present in the washing solution. The washed gels were stained with the Coomassie Blue, dried and radioautographed. The radioautographs and dried protein-stained gels (immersed in chloroform) were scanned at 550nm. The bar in the absorbance scale of each Figure represents an absorbance increment of 1.0 for the radioautograph scans, and an absorbance increment of 0.5 for the protein-stain scans. In both Figures the scans are of: (i) protein-stained gel; (ii) radioautograph of gel incubated with 1251_ labelled concanavalin A without inhibitor; (iii) radioautograph of gel incubated with 125I-labelled concanavalin A in the presence of methyl a-mannoside.

490

D. G. WILLIAMS, R. E. JENKINS AND M. J. A. TANNER (a) AB

D

C

E

4,5 N

Ni6,7,11,12,13

14

1,2

Thl1

3/' 8,9,10 I

IC T

T Th3

-

Th4 1 Th5l ThT |

)Th6 T T5s1 IT6s T4 T5s -

T2s- -

Th

(b) Th DIDS I

-*

2

314

Lectin receptors 4,5 /T

5\\

Out

'C

1

,

N

13

Fig. 8. Alignment of the fragments and topography of the anion-transport protein

(a) Alignment of the fragments on the amino acid sequence. The numbers refer to the 'l25-labelled sites, which yield the corresponding labelled peptides on peptide 'maps' of thermolysin (Th) and trypsin (T) digests of the protein. The broken lines indicate alternative positions for certain sites and fragments. The thickened portions of the line for the intact protein represent extracellular regions, and the upper-case letters show the positions of the loops referred to in (b). The suffix s is used to indicate water-soluble fragments. (b) Schematic diagram of the folding of the polypeptide chain in the membrane. The broken lines show alternative positions for the labelled sites in the C-terminus of the molecule and alternative possibilities for the topography of the protein in this region. The diagram shows the major (solid arrow) and minor (broken arrow) sites of cleavage by thermolysin (Th) and of trypsin (T), and the regions that bind DIDS and lectins. The detailed evidence f6r this model is given in the text and is briefly summarized here. The N-terminus of the molecule is contained in the large fragment T2s, which is located in the cytoplasmic domain of the protein. Since the N-termini of fragments Th4 and Th5 result from minor extracellular cleavages by thermolysin (at loop A) the polypeptide traverses the membrane between the C-terminus of fragment T2s and the N-terminus of fragment Th4. The Nterminus of fragment ThT arises from cleavage by trypsin at the cytoplasmic loop B, but the C-terminus of this fragment is produced by the major extracellular thermolysin cleavage at loop C. The fragment Th6 results in a minor thermolysin cleavage in loop E.

125I-labelled sites in the molecule. Fragment Thl extends from the N-terminus of the molecule and has tyrosine as a C-terminal residue (Drickamer, 1976; Jenkins & Tanner, 1977b) and this tyrosine residue probably gives rise to the labelled peptide 3 on the peptide 'maps'. Fragment Th3 contains the remaining '25I-labelled sites that are present in the intact molecule and not present in fragment Thl. Thus fragment Th3 overlaps fragment TI, the major product of intracellular cleavage of the protein by trypsin. Fragment Ti probably extends from site 14 to the C-terminus of the molecule and like fragment Th3 contains sites 4 and 5 and sites 8, 9 and 10 which are not present in fragment Thl (Jenkins & Tanner, 1975, 1977a). As would be expected, the oligosaccaride receptors for concanavalin A and the R. communis lectin are present in both fragments Th3 and TI, but are not present in fragment Thi (Jenkins & Tanner, 1977b). The minor fragment Th6 contains only the extracellular sites 4 and 5 and is clearly a subfragment of fragment Th3. The appearance of this fragment and the observation that labelled peptides containing sites 4 and 5 are released from erythrocytes on thermolysin treatment (Jenkins & Tanner, 1975) probably accounts for the low and variable yield of fragment Th3 and the equivalent fragments produced by other proteinases (Steck et al., 1976). Fragment T6s, the very small fragment that is solubilized by extensive trypsin treatment of erythrocyte 'ghosts' and contains only the intracellular sites 8, 9 and 10, is also derived from the sequence in fragment Th3. These sites 8, 9 and 10 are present in the C-terminal portion of the molecule. The relative order of t-he two groups of sites 4 and 5 and sites 8, 9 and 10 cannot yet be defined. Fragment Th3 corresponds to the labile fragment of apparent mol.wt. about 40000, which is obtained in addition to fragment Thl on treatment of erythrocytes with Pronase (Cabantchik & Rothstein, 1974b) and chymotrypsin (Steck et al., 1976, 1978). The polypeptide in this fragment from the C-terminus of the molecule is clearly acessible from both sides of the membrane, since it contains both intracellular and extracellular labelled sites. Fragments Th4, Th5 and ThT are all derived from the same region of the protein, that is the C-terminal region of fragment Thl and share the C-terminus of fragment Thl. Fragments Th4 and Th5 differ from If fragment Th6 is the C-terminal peptide then the cleavage occurs on the N-terminal side of sites 4 and 5 and sites 8, 9 and 10 are on the cytoplasmic loop D (shown by the solid line). Alternatively the cleavage may occur on the C-terminal side of sites 4 and 5, in which case sites 8, 9 and 10 are present on a cytoplasmic 'tail' at the C-terminus of the protein (shown as a broken line).

1979

STRUCTURE OF ERYTHROCYTE MEMBRANE ANION-TRANSPORT PROTEIN fragment ThT in that they are products of extracellular proteolysis and they both contain site 14, which is located in the cytoplasmic domain of the protein. Fragment ThT does not contain site 14 and the trypsin cleavage leading to the N-terminus of fragment ThT occurs in the intracellular domain of the protein. Fragment TI, the product of trypsin cleavage of the intact protein from the cytoplasmic side of the membrane, also lacks site 14 (Jenkins & Tanner, 1975, 1977a). It is likely that the trypsin cleavages leading to the N-termini of both fragments ThT and TI occur at the same point in the cytoplasmic domain on the C-terminal side of site 14. The alignment of Fig. 8(a) suggests that trypsin digestion of leaky membranes should yield a small membrane-bound fragment derived from the region between the C-terminus of fragment T2s and the N-terminus of fragment T1. Because of uncertainty of the exact molecular weight of fragment Thl (estimates of the apparent molecular weight of this fragment range from 55000 to 70000; see Cabantchik et al., 1978) it is difficult to predict the size of this fragment. It is likely to be a relatively small hydrophobic fragment that would make it difficult to detect and may not be labelled if site 14 is solubilized from the membrane. No fragment containing site 14 was identified in the trypsin-solubilized material from the anion-transport protein. However, the electrophoretic mobility of labelled peptide 14 suggests that the labelled tyrosine residue of site 14 is in a rather basic peptide. lt is possible that trypsin cleavage occurs within the sequence in peptide 14, giving rise to a new labelled peptide with a different mobility on peptide 'maps'. In previous work we obtained a band of apparent mol.wt. 24000 (fragment T2) on trypsin digestion of the anion-transport protein in leaky erythrocyte 'ghosts' (Jenkins & Tanner, 1 977a). (This band should not be confused with the soluble fragment designated T2s in the present study.) The T2 band contained the extracellular sites 1, 2, 3, 4 and 5, as well as the intracellular sites 6, 7, 11, 12 and 13, and was assumed to be a homogeneous polypeptide derived from the N-terminus of the protein, which contained the portion of the polypeptide that was not present in the major tryptic fragment TI. In this previous work the fragments were separated by electrophoresis on gels of uniform acrylamide concentration. In the present work we have used gradient-gel separations that give a greatly improved resolution of low-molecular-weight material. A band containing the same labelled site as fragment T2 was obtained in two separate experiments, but in very low yield. One explanation for these results is that the band T2 contained more than one polypeptide. It could contain the soluble fragment T3s (which has mol.wt. 23000) co-migrating on gel electrophoresis with a minor fragment containing the extracellular labelled Vol. 181

491

site. Unsealed erythrocyte 'ghosts' tend to form sealed right-side-out vesicles on proteolysis (Avruch et al., 1973; Steck et al., 1976) and traces of the fragment T3s may have become entrapped within the membrane. It has been clearly established that a region of approximate mol.wt. 40000 containing the N-terminus of the protein is totally in the aqueous domain at the cytoplasmic side of the membrane (Steck et al., 1976, 1978; Drickamer, 1978; Fukuda et al., 1978); thus if band T2 contained a single membrane-bound polypeptide this peptide could not extend to the N-terminus of the protein. However, the observation that the extracellular sites 2 and 3 present in band T2 have different characteristics of iodination with lactoperoxidase from the sites 2 and 3 present in fragment TI remains unexplained (Jenkins & Tanner, 1977a). The order of the labelled sites on the protein sequence (Fig. 8a), together with their known location in the extracellular or intracellular portion of the protein, show that from the N-terminus, which is located on the cytoplasmic side of the membrane, the polypeptide penetrates the membrane to the extracellular surface and returns to the cytoplasmic side of the membrane at some point in the C-terminal region of the protein. It is not known whether the C-terminus of the molecule is located at the intracellular or extracellular side of the membrane. This structure is similar to the models proposed by Steck et al. (1978) and Grinstein et al. (1978), except that we have been able to show that a section in the C-terminal third of the polypeptide is also part of the cytoplasmic domain of the protein. A more detailed model for the topography of the protein can be derived by including the locations of the sites of proteolytic cleavage that give rise to the termini of the fragments (Fig. 8b). A domain extending approx. 40000 daltons from the N-terminus of the protein is present at the cytoplasmic surface of the membrane (Steck et al., 1978; Fukuda et al., 1978) and gives rise to fragment T2s. Since the Ntermini of fragments Th4 and Th5 result from extracellular cleavage, the polypeptide chain must penetrate the membrane from the intracellular to the extracellular surface at the C-terminal side of fragment T2s. The polypeptide chain then forms loop A (Fig. 8b) and returns to the cytoplasmic side of the membrane since the N-terminus of fragment ThT results from cleavage at this side of the membrane and fragment ThT lacks the intracellular site 14, whereas fragments Th4 and Th5 both contain this site. Since fragments Thl, Th4, Th5 and ThT share a common C-terminus derived from extracellular cleavage around site 3, the chain again traverses the membrane to the extracellular surface forming loop B. Fragment Th3 (and fragment TI; see Fig. 8a) contain both the extracellular sites 4 and 5 and the intracellular sites 8, 9 and 10. Thus the region of the

492


protein containing fragment Th3 must also traverse the membrane from the extracellular to the cytoplasmic side of the membrane on the C-terminal side of site 3 to form loop C. The membrane-bound fragment Th6 contains only the extracellular sites 4 and 5, and is also derived from this region of the molecule by extracellular cleavage. The absence of the intracellular sites 8, 9 and 10 from this fragment requires that the polypeptide chain makes a further extracellular loop (E) out of the membrane near its C-terminal end. There are two possible structures for this region of the molecule depending on the position of the groups of sites 4 and 5 and sites 8, 9 and 10. In one (shown by the solid line in Fig. 8b) sites 8, 9 and 10 are on the intracellular loop D, whereas sites 4 and 5 are on the C-terminal extracellular loop E. In this case the C-terminus of the protein may be imbedded in the membrane and does not necessarily penetrate through the membrane to the cytoplasmic site. In the alternative model (depicted by a broken line in Fig. 8b) sites 8, 9 and 10 are present on an intracellular tail at the C-terminus of the molecule. Here the polypeptide need not necessarily penetrate the membrane to form loop D, but this segment of the protein could simply be imbedded in the membrane. In this case the extracellular sites 4 and 5 could either be on loop C (adjacent to site 3) or loop E, depending on the position of the cleavage leading to fragment Th6. The presence of binding sites for concanavalin A and the lectin from R. communis in fragment Th3 confirms the location of these receptors in the Cterminal region of the protein, in an extracellular loop (C or E) C-terminal to the point of cleavage leading to fragment Th . Drickamer (1977) suggested that the polypeptide may cross the membrane four times. His results with a different group of labels and chemical fragmentation of the isolated protein are consistent with the presence of loops A, B and C in the protein (Fig. 8b). Drickamer (1977) calculated the distribution of hydrophobic amino acids in different regions of the anion-transport protein and the regions of high-hydrophobic amino acid content correspond in position to the membrane-traversing regions that we have found. The detailed model in Fig. 8b is based on assumptions about the sites of proteolytic cleavage. Whereas the major sites of proteolytic cleavage (leading to fragments Thi and T2s) are well characterized (see reviews by Cabantchik et al., 1978; Steck, 1978), the sites yielding the minor fragments are less certain. The conclusion that the N-termini of fragments Th4 and Th5 and the termini of fragment Th6 result from extracellular cleavage is based on the occurrence of these fragments on thermolysin digestion of erythrocytes and resealed 'ghosts' and it is difficult to obtain confirmatory evidence that this is indeed the case. Since these are minor fragments

it is conceivable that a small proportion of the erythrocytes became leaky during the digestion and the proteinase acted in part at the intracellular surface. However, the extensive washing of the cells at low centrifugal forces would be expected to remove the lysed cells, which require much higher centrifugal forces for sedimentation. In addition, unpublished experiments have shown that thermolysin treatment of erythrocyte 'ghosts' yields a fragment with the same gel-electrophoretic mobility and labelled peptide 'map' as fragment ThT and no fragments with a mobility corresponding to fragments Th4 or Th6 were obtained. Similarly ifthe region contained in the glycosylated fragment Th3 contained only a single traverse of the polypeptide from the extracellular region near site 3 to the cytoplasmic side of the membrane and fragment Th6 were derived from intracellular cleavage, one would expect fragment Th6 to migrate as a diffuse band rather than the well defined band shown in Fig. 3(e). In this case, fragment Th6 would be expected to extend from the C-terminal side of site 3 and contain the glycosylated region of the protein. This glycosylation appears to be responsible for the diffuse bands given by the intact protein and fragments TI and Th3 (Jenkins & Tanner, 1977b). The observed result is consistent with the presence of the extracellular loop E in Fig. 8(b). Two points should be made about the model shown in Fig. 8(b). First, the model shown is the simplest one that accommodates the experimental results reported in the present paper. Since the techniques we have used can yield only a minimum estimate of the number of traverses of the polypeptide chain across the membrane, further experimental results may require a model with additional folds of the polypeptide chain in certain regions. Secondly, the structure shown is highly diagrammatic to illustrate the general topography of the protein. It is not intended to represent the threedimensional structure of the protein. It is likely that the membrane-traversing segments of the protein are organized to form an ordered structure perhaps of the type found in the bacterial proton pump, bacteriorhodopsin (Henderson & Unwin, 1975), which contains seven membrane-traversing segments. This type of structure could be responsible for some of the transport properties of the protein by forming a channel through the membrane. We thank Diana Barker, Sally Birch and Simon Peake for their assistance at certain stages of this work. R. E. J. and D. G. W. were recipients of Science Research Council Studentships. The work was supported by a grant from the Wellcome Trust.

References Avruch, J., Price, H. D., Martin, D. B. & Carter, J. R. (1973) Biochim. Biophys. Acta 291, 494-505

1979

STRUCTURE OF ERYTHROCYTE MEMBRANE ANION-TRANSPORT PROTEIN Bender, W. W., Garan, H. & Berg, H. C. (1971) J. Mol. Biol. 58, 783-797 Bodeman, H. & Passow, H. (1972) J. Membr. Biol. 8, 1-26 Boxer, D. H., Jenkins, R. E. & Tanner, M. J. A. (1974) Biochemn. J. 137, 531-534 Bretscher, M. S. (1971) J. Mol. Biol. 59, 351-357 Cabantchik, Z. I. & Rothstein, A. (1974a) J. Membr. Biol. 15, 207-226 Cabantchik, Z. I. & Rothstein, A. (1974b) J. Membr. Biol. 15, 227-248 Cabantchik, Z. I., Knauf, P. A. & Rothstein, A. (1978) Biochim. Biophys. Acta 515, 239-302 Dodge, J. T., Mitchell, C. & Hanahan, D. J. (1963) Arch. Biochem. Biophys. 100, 119-130 Drickamer, L. K. (1976)J. Biol. Chem. 251, 5115-5123 Drickamer, L. K. (1977) J. Biol. Chem. 252, 6909-6917 Drickamer, L. K. (1978) J. Biol. Chem. 253, 7242-7248 Fairbanks, G., Steck, T. L. & Wallach, D. F. H. (1971) Biochemistry 10, 2606-2617 Findlay, J. B. C. (1974) J. Biol. Chem. 249, 4398-4403 Fukuda, M., Eshdat, Y., Tarone, G. & Marchesi, V. T. (1978) J. Biol. Chem. 253, 2419-2428 Grinstein, S., Ship, S. & Rothstein, A. (1978) Biochim. Biophys. Acta 507, 294-304 Henderson, R. & Unwin, P. N. T. (1975) Nature (London) 257, 28-32 Ho, M. K. & Guidotti, G. (1975) J. Biol. Chem. 250, 675-683 Jenkins, R. E. & Tanner, M. J. A. (1975) Biochem. J. 147, 393-399

Vol. 181

493

Jenkins, R. E. & Tanner, M. J. A. (1977a) Biochem. J. 161, 131-138 Jenkins, R. E. & Tanner, M. J. A. (1977b) Biochem. J. 161, 139-147 Laemmli, U. K. (1970) Nature (London) 227, 680-682 Lepke, S. & Passow, H. (1976) Biochim. Biophys. Acta 455, 353-370 Lepke, S., Fasold, H., Pring, M. & Passow, H. (1976) J. Membr. Biol. 29, 147-177 Passow, H. (1969) in Laboratory Techniques in Membrane Biophysics (Passow, H. & Stampfli, R., eds.), pp. 21-27, Springer-Verlag, Berlin Passow, H., Fasold, H., Lepke, S., Pring, M. & Schuhmann, B. (1977) in Membrane Toxicity (Miller, M. W. & Shamoo, A. E., eds.), pp. 353-379, Plenum Press, New York Ship, S., Shami, Y., Breuer, W. & Rothstein, A. (1977) J. Membr. Biol. 33, 311-323 Steck, T. L. (1974) J. Cell. Biol. 62, 1-19 Steck, T. L. (1978) J. Supramol. Struct. 8, 311-324 Steck, T. L., Ramos, B. & Strapazon, E. (1976) Biochemistry 15, 1154-1161 Steck, T. L., Koziarz, J. J., Singh, M. K., Reddy, G. & Kohler, H. (1978) Biochemistry 17, 1216-1222 Tanner, M. J. A. & Anstee, D. J. (1976) Biochem. J. 153, 265-270 Tanner, M. J. A., Jenkins, R. E., Anstee, D. J. & Clamp, J. R. (1976) Biochem. J. 155, 701-703 Yu, J. & Steck, T. L. (1975a) J. Biol. Chem. 250, 91709175 Yu, J. & Steck, T. L. (1975b) J. Biol. Chem. 250, 91769184